791 results about "Relation graph" patented technology

The invention provides a space field game behavior dynamic evolution and strategy deduction optimization method and system, and relates to the technical field of space field game behavior optimization, and the method comprises the steps: analyzing historical behavior data through a deep neural network, constructing a dynamic relation graph of participants, inputting a multi-agent reinforcement learning model, and outputting a game behavior model. Then, a multi-layer game scene evolution tree is constructed, the instant income and the long-term risk are calculated in combination with a dual neural network, dynamic updating is performed by using a search algorithm of a hierarchical attention mechanism, and candidate evolution paths are generated; and finally, evaluating the candidate paths by using a strategy evaluation model constructed by a space-time diagram neural network, selecting an optimal strategy path, and generating a strategy deployment scheme implemented by stages in combination with a knowledge inference engine. According to the method, the evolution trend of the game behavior in the space field can be effectively predicted, a feasible strategy deployment scheme is provided, and the scientificity and effectiveness of strategy decision are improved.

According to the research and development document processing method and device provided by the embodiment of the invention, unified processing of text, voice and image information is realized by constructing the multi-mode document analysis engine. Different types of data are converted into standardized feature representations through cross-modal preprocessing and a semantic alignment network. The system adopts a deep learning model to identify research and development elements, establishes a multi-modal relation graph, and realizes intelligent extraction and correlation analysis of various types of information in research and development documents. According to the method, the defects of the traditional technology in the aspects of multi-modal information fusion and knowledge system construction are effectively overcome, powerful support is provided for research and development process management and knowledge asset accumulation, and the standardized management level of research and development documents is remarkably improved.

The invention discloses an intelligent target allocation method and system based on deep reinforcement learning, particularly relates to the field of multi-agent systems, and is used for solving the real-time decision problem under multi-dimensional constraints in a dynamic confrontation simulation scene. Time, space and resource constraints are decomposed into independent models, and the decision complexity of a multi-agent system in a dynamic confrontation scene is reduced; by means of adaptive Monte Carlo sampling, prediction perception path planning and dynamic threshold value resource allocation, diversified high-quality candidate schemes are quickly generated; when the environment changes, the multi-layer influence propagation network and the spatial topological relation graph are used for positioning an affected agent, and scheme fragments are generated through local adjustment, so that global re-calculation is avoided, the overhead is reduced, and real-time response is ensured; meanwhile, an efficient and reliable optimal solution is screened out through resource delay coupling and path efficiency comprehensive evaluation in combination with multi-objective optimization and task consistency verification.

The invention provides an industrial chain atlas modeling method and system based on deep learning, and relates to the technical field of industrial chain analysis and modeling, and the method comprises the steps: collecting multi-source heterogeneous data in an industrial chain, including demand information, resource states, process features and operation and maintenance data, and preprocessing the multi-source heterogeneous data to form a standardized data set; and performing semantic vectorization processing on the text type data in the standardized data set, and performing node embedding on the multi-relation graph to generate entity low-dimensional representation fusing text and structural features. According to the method, by integrating multi-source heterogeneous data, fusing text and structural features, constructing an industrial chain knowledge graph and performing dynamic optimization, accurate adjustment of production line configuration and task allocation strategies is realized, meanwhile, the industrial trend is effectively predicted, risk paths are analyzed and resource bottlenecks are identified in combination with a risk propagation model, a scientific basis is provided for industrial chain decision-making, and the risk propagation model has a wide application prospect. And the operation efficiency and stability of an industrial chain are improved.

The invention relates to the technical field of archive information management, and provides an archive information extraction management method and system based on multi-modal learning, and the method comprises the steps: collecting multi-modal archive data, carrying out the preprocessing of the multi-modal archive data, and obtaining the processed multi-modal archive data; performing feature extraction on the multi-modal archive data, and obtaining an archive comprehensive feature vector through weight distribution and feature alignment; based on a preset archive knowledge graph, performing multi-modal entity recognition on the archive comprehensive feature vector, constructing an archive relation graph, and performing iterative calculation through a graph neural network to obtain an archive association rule; updating a preset archive knowledge graph according to the archive association rule to obtain an archive enhanced knowledge graph, and performing iterative optimization on the archive association rule to form and store an archive management strategy network; and constructing an archive information index system based on the archive enhanced knowledge graph and the archive management strategy network. According to the invention, efficient management and accurate retrieval of archive information are realized.

The invention provides a private domain social relation mining method and system based on a graph neural network, and relates to the technical field of social relation mining, and the method comprises the steps: carrying out the entity alignment and relation mapping of historical behavior data of a private domain platform user and an external knowledge base through employing a knowledge graph technology, generating user behavior feature data, and constructing a user relation graph; processing the user data by using a federal map neural network model containing a space-time attention layer to obtain a user node representation vector; and according to the user node representation vector, calculating a social influence index and a relationship strength index between the users, determining a final social circle layer structure by adopting a spectral clustering method and causal intervention analysis, and generating a social relationship analysis report. According to the method, complex user social relations in a private domain platform can be effectively mined, the accuracy and stability of social circle layer identification are improved, and technical support is provided for precise marketing and personalized recommendation.

The invention discloses an energy consumption model intelligent construction system and method based on artificial intelligence, and relates to the technical field of artificial intelligence, and the system comprises an Internet of Things multi-source data collection module, a data cleaning and space-time calibration module, a multi-source data semantic fusion module, a dynamic energy consumption relation graph construction module, an intelligent decision engine module and an edge-cloud collaborative deployment module. According to the method, multi-source data are fused through a Transform multi-head self-attention mechanism, a dynamic energy consumption relation graph is constructed by using a graph neural network, and dynamic modeling and intelligent regulation and control of energy consumption are realized in combination with an edge-cloud hierarchical decision architecture; the method comprises the steps of data acquisition and standardization, cleaning calibration, semantic fusion, graph modeling, hierarchical decision making and collaborative execution. According to the method, the problems of insufficient data integration and model staticization of a traditional system are solved, the accuracy, real-time performance and global optimization capability of energy consumption management are improved, the method is suitable for scenes such as intelligent buildings, the energy efficiency is remarkably improved, and the data security is guaranteed.

The invention relates to the technical field of point cloud data modeling, and discloses a point cloud data processing and modeling method, system and device for power distribution equipment and a storage medium, and the method comprises the steps: collecting the point cloud data of the power distribution equipment, and constructing a topological relation graph according to the topological relation of the power distribution equipment; semantic segmentation is performed on the point cloud data based on the topological relation graph, a component association model is constructed, and missing data detection is performed in combination with the topological relation graph; and generating a high-precision three-dimensional model based on the component correlation model. According to the method, semantic segmentation is guided through the topological relation graph, the accuracy of point cloud data classification is improved, and the problem of misclassification in a traditional method is avoided. Meanwhile, a part correlation model is generated based on the segmentation result, the geometric and functional connection relation between the devices is comprehensively described, and deep fusion of the device connection information and the three-dimensional geometric model is achieved.

The invention provides a multi-robot collaborative task dynamic scheduling method and related equipment, which are applied to the technical field of robots, and the method comprises the following steps: collecting task scheduling scene information; the task scheduling scene information is used for constructing a multi-level graph structure, multi-dimensional input information is obtained based on the multi-level graph structure, the multi-level graph structure comprises a relation graph layer of multiple dimensions, and the relation graph layer comprises relation information between objects of the dimensions corresponding to the relation graph layer; determining a task allocation strategy based on the multi-dimensional input information; and based on the task allocation strategy, sending a task execution instruction to each target robot, and controlling the target robot to execute a task corresponding to the task execution instruction. Complicated relations among the robot, the environment and the tasks are expressed by using the multi-level graph structure, rich and complete relation information is provided for generating the task allocation strategy, the accuracy of the generated task allocation strategy is improved, and conflicts generated during task allocation are avoided.

The invention relates to the technical field of defect detection, in particular to a cross-process defect root cause tracing method and system. According to the method, the data feature matrix covering multiple dimensions is formed by integrating the process parameters, the equipment state and the quality detection information, so that the performance evaluation of each process is more comprehensive, the interaction and influence paths among the processes can be clearly described by constructing the process relation graph, and the performance evaluation efficiency is improved. Meanwhile, basic data support is provided for quantifying the relation between the procedures through introduction of procedure attenuation factors, the shortest propagation path and the propagation probability of the defects can be accurately recognized by analyzing a procedure relation graph, root cause tracing of the cross-procedure defects becomes systematized in combination with construction of a knowledge graph, and the defect tracing efficiency is improved. The knowledge graph not only can effectively integrate and display data, but also is convenient for quickly positioning problems, and by utilizing an adaptive correlation analysis technology, the system can intelligently adjust an analysis model and a path and continuously optimize a defect detection and tracing process when facing new data.

The invention provides an abnormal behavior recognition method and system based on multi-dimensional data analysis, and the method comprises the steps: carrying out the cleaning and standardization processing of structured data and unstructured data, obtaining a data set in a unified format, employing a natural language processing technology, processing the document text in the data set, extracting the entity, relation and behavior information, and carrying out the recognition of abnormal behaviors. Generating a structured feature vector; constructing behavior statistics, time sequence and association features by using the feature vectors and the structured data, selecting a feature combination by adopting a sampling algorithm, establishing an entity relation graph according to association data in a constructed multi-dimensional feature set, calculating association strength among entities, and optimizing the entity relation graph according to the association strength to establish a network structure model; a feature weight system is set according to the model and a multi-dimensional feature set, a multi-dimensional abnormal score is calculated by using a pruning tensor structure measurement method, an abnormal score is comprehensively calculated by using a low-rank tensor recovery technology, and an early warning list is output, so that the recognition accuracy and efficiency can be improved.

The invention relates to the technical field of artificial intelligence, and discloses an anti-fraud model training method based on deep learning, which comprises the following steps: step 1, preprocessing and constructing multi-modal data, carrying out abnormal value processing, missing value filling and feature standardization on original data, representing processed transaction data and geographic information through a feature extraction module, and carrying out feature extraction; the equipment information and the social relation graph are embedded through an embedding layer and a graph neural network; step 2, based on data enhancement of self-supervised learning, pre-training unlabeled data by designing a behavior prediction task and a mask filling task so as to initialize a main model by using generated model parameters, and generating a synthetic sample extension training data set; by designing a behavior prediction task and a mask filling task, automatic learning and enhancement of features are realized by using label-free data, the dependence on labeled data is reduced, the detection capability of the model on fraudulent behaviors is remarkably improved, and the diversity of training data is enriched.

The invention provides a digital asset-oriented network security management method and system. The method comprises the following steps: acquiring a digital asset list and a business process document, identifying keywords in the document by using a natural language processing technology, constructing an association relationship graph of digital assets and business processes, and discovering potential association; directly connected business process nodes of each digital asset are extracted from the incidence relation graph, an asset business influence range report is generated, and a many-to-many mapping relation between the assets and the businesses is displayed in a matrix form; in the review process, association and quantitative indexes, including a risk thermodynamic diagram, an influence degree scatter diagram and the like, among assets, risks and business processes are dynamically presented through a visualization technology, and a visual information transmission channel is provided for all parties; and after meeting review is consistent, qualitative influence analysis of the business department and quantitative loss estimation of the technical department are integrated into a business influence analysis result by adopting a fuzzy comprehensive evaluation method, and a comprehensive business influence level is obtained.

The invention relates to a water quality prediction method and system based on gating residual enhancement and feature fusion, and belongs to the technical field of water environment intelligent analysis and deep learning. Taking each water quality index as a node of the graph, and constructing two complementary variable relation graph structures by utilizing a Pearson's correlation coefficient and mutual information; respectively inputting the two graph structures into a graph convolutional network, extracting deep dependency features among indexes, and splicing and fusing the deep dependency features. A multi-head attention mechanism is used as a trunk to extract global time dependence, a GRU network is introduced to extract local time sequence features, GRU output is used as an adjustable residual term to be injected into the attention trunk through a residual gating mechanism, self-adaptive enhancement of local dynamic features is achieved, and finally a self-adaptive fusion mechanism is introduced to generate comprehensive representation. According to the method, the complex dependency relationship between the water quality indexes and the time dynamic evolution process can be modeled in a collaborative manner, the response capability to key local change and sudden change events is remarkably enhanced, and the accuracy and robustness of water quality prediction are improved.

The invention relates to the technical field of creative design, in particular to an intelligent creative method and system based on design elements, and the method comprises the following steps: collecting design element data, analyzing dependency and influence among elements through a graph network, extracting element attributes, calibrating dependency intensity, and carrying out element association mapping to obtain a design element relation graph. According to the method, analysis dimensions of design data are enhanced through deep utilization of a graph network, deep insight of complex relations among design elements is provided, element dependence and influence are evaluated in a quantitative mode, data driving performance of a decision process is ensured, and sentiment analysis is carried out in combination with a design element relation graph. The design scheme can reflect user emotion requirements in detail, design template updating dynamically adapting to market changes, design iteration acceleration and reaction speed and market competitiveness improvement are achieved, the effectiveness of the scheme is ensured through design evaluation and analogue simulation of user experience, and the market release risk is reduced.

The invention discloses a cross-modal alignment method and device for a multi-modal large language model, and relates to the technical field of multi-modal alignment, and the method comprises the steps: obtaining an image, carrying out the target detection, and generating a local image fine-grained feature; obtaining a text, extracting text features, extracting a text entity relationship, and obtaining an entity relationship dependency tree as a text entity relationship graph; performing local fine-grained semantic alignment on the local image fine-grained features and the text features; using the scene graph model to obtain a visual relation graph between detection targets in the image; performing graph structure alignment and graph semantic alignment on the text entity relation graph and the visual relation graph; and fusing the semantic alignment features and the graph alignment features to obtain semantic alignment features with rich fine grit, global semantic alignment features and richer relationship alignment features, and inputting the semantic alignment features, the global semantic alignment features and the richer relationship alignment features into a multi-modal large model. According to the method, alignment of different modal data on different feature dimensions is realized, so that the different modal data are accurately associated.

The invention provides a project management software document uploading method based on OCR and a large language model, and relates to the technical field of large language models, and the method comprises the steps: receiving a project document picture, and carrying out OCR recognition through a multi-scale feature pyramid and an attention-enhanced convolutional neural network; inputting the text information into a pre-trained large language model for classification and key field extraction; determining a data partitioning strategy according to the item type to create a storage space; establishing a multi-dimensional index structure to store project information and analyzing a content logic relationship; and generating and displaying a project information relation graph. According to the invention, intelligent identification, classification and correlation analysis of the project documents are realized, and the project management efficiency and the data utilization value are improved.

The invention relates to the technical field of database management, in particular to a two-way linkage database table and supervision submission form field synchronous construction method which is applied to a supervision data submission scene. According to the scheme, the method comprises the steps that physical structure metadata of a database table and definition metadata of a supervision submission form are obtained, a semantic vector set is formed by combining metadata semantic analysis and semantic coding, and a bidirectional mapping relation graph is constructed through relation weaving; constructing a linkage propagation path according to a field change event, realizing adaptive bidirectional structure adjustment, generating a field synchronous construction scheme, and driving bidirectional mapping relation graph evolution optimization through feedback learning; according to the method, the bidirectional mapping relation graph and self-adaptive bidirectional structure adjustment are creatively combined, and efficient and self-adaptive bidirectional field synchronous construction is realized.

The invention discloses a park management method based on digital twinning, particularly relates to the field of park operation, and is used for solving the problem that a multi-source event is difficult to stably chain under the conditions of compensation delay and evidence gap and causes misalignment of disposal triggering. Comprising the following steps of: registering an event in a pool to generate an event number and writing the event number into a platform arrival moment; screening and mapping an anchor point event to a park object identifier and a process stage identifier to generate a management token; checking a stage sequence and an evidence item requirement set according to an event chain rule table in event chain construction and generating a conflict point list; the delay portrait library generates a credible mark when a source is generated and forms a reverse sequence violation spectrum; the digital twin object relation graph forms an influence domain span mark; the closed table is returned to obtain a disposal convergence guarantee level; the cost-sensitive decision tree outputs a disposal template identifier, and the disposal template table generates a management action instruction and a trigger voucher and records an instruction acceptance identifier; and returning, comparing and freezing the exception management token, and updating the event chain rule table and the delay portrait library.

The invention relates to the technical field of databases, and discloses a multi-type database performance optimization and operation and maintenance management method and system.The multi-type database performance optimization and operation and maintenance management method comprises the steps that real-time performance indexes of a heterogeneous database cluster are collected, and a noise reduction data set is obtained; generating a cross-library performance coupling degree matrix through the resource competition coupling degree and the data dependence coupling degree; constructing a coupling relation graph; identifying a bottleneck node set based on the coupling relation graph and the abnormal level; generating an optimization strategy; and performing strategy conflict identification according to the optimization strategy, generating a processing strategy, and executing the processing strategy optimization strategy. Through the improved weighting centrality algorithm, the core bottleneck node which has the greatest influence on the whole cluster is accurately identified, a differential optimization strategy is adopted according to the database type, a perfect strategy conflict detection and avoidance mechanism is established, mutual interference in the optimization process is effectively prevented, the optimization success rate is improved, and the optimization efficiency is improved. And the optimization time is shortened.

The invention, which relates to the technical field of ecological environment monitoring and intelligent management, discloses a real-time monitoring and early warning management system based on an ecological environment, comprising a multi-modal data acquisition module for acquiring and preprocessing multi-medium environment data, a multi-medium risk conduction modeling module for constructing a directed graph, and a dynamic modeling risk conduction process. The cross-medium risk conduction chain effect evaluation system has the advantages that the problem of cross-medium risk conduction chain effect evaluation deficiency is effectively solved, multi-modal data provides a data basis, the modeling module constructs a directed conduction relation graph, and the model construction module constructs a three-dimensional model and simulates pollutant migration and transformation and risk effects, and the intelligent early warning module generates early warnings, so that the cross-medium risk conduction chain effect evaluation system has the advantages that the cross-medium risk conduction chain effect evaluation deficiency problem is effectively solved; a graph neural network is used for dynamic modeling, a conduction path is presented, a digital twin engine simulates pollutant migration and transformation and a risk conduction effect, an intelligent early warning module carries out multi-level evaluation and timely early warning, a decision intervention module generates an optimal intervention scheme, and the system realizes accurate evaluation and effective response of cross-medium risk conduction.

The invention discloses an electric power production data private network security situation awareness system and method. The method comprises the following steps: 1) acquiring security data of an electric power production data private network in real time; 2) standardizing the acquired data; 3) constructing a power grid entity relation graph, and fusing the standardized data obtained in the step 2) into the graph; 4) constructing an attack graph, and associating the attack graph with the power grid equipment entity relationship graph; 5) combining the power grid entity relation graph with the attack graph, performing deep correlation analysis, and identifying potential attacks in the power grid; 6) generating a grade risk alarm according to an identification result in the step 5), and triggering corresponding countermeasures; and 7) simulating the defense measures in the step 6), and optimizing and adjusting the defense countermeasures according to a feedback result. According to the method, the multi-source data is combined with the dynamic knowledge graph technology, real-time correlation analysis of the power grid entity relationship and the attack behavior is supported, and the recognition capability of novel attacks is remarkably improved.

The invention relates to the technical field of data processing and analysis, and discloses a cost consultation data arrangement and analysis system based on a cloud platform, and the system comprises a data collection module which is used for obtaining and standardizing original cost data; the data clustering module is used for clustering data according to specific dimensions and extracting key feature vectors; the strategy generation module is used for constructing candidate analysis strategies; the strategy optimization module is used for screening an optimization strategy; and the execution engine module processes the data according to the optimization strategy to generate an analysis result. In addition, a data verification module is arranged to verify results, a dynamic adjustment module is arranged to monitor data changes, and a knowledge graph module is arranged to construct an entity relation graph. The system can solve the problems of data source heterogeneity, single analysis strategy and the like in cost data processing, realizes efficient data processing and accurate analysis, ensures that the result is accurate and reliable, and improves the data processing and analysis level of the cost consultation industry.

The invention relates to the technical field of data fusion, and discloses an animal scene-oriented adaptive multi-modal data fusion method, which comprises the following steps of: extracting spatio-temporal characteristics from multi-source heterogeneous data such as visual sense, auditory sense and physiological sensing, constructing an animal-environment-group ternary spatio-temporal relation graph, and constructing an animal-environment-group ternary spatio-temporal relation graph; a pilot frequency sampling problem is solved through an adaptive interpolation algorithm, cross-modal projection alignment is completed in a public semantic space, unified space-time representation is output, and confidence coefficient weight is dynamically calculated based on uncertainty measurement of each modal feature. According to the method, accurate alignment of multi-modal data is realized through the cross-modal space-time attention network, the multi-modal feature alignment error is reduced compared with that of a traditional LSTM method, the training data volume of a federated element migration reinforcement learning framework is reduced compared with that of a traditional migration learning method, and the cross-species generalization performance of the model is improved. A multi-level causal inference engine quantitatively reveals causal association between environmental factors and animal diseases, and in combination with a dynamic decision tree visualization technology, the decision recognition degree is improved.

The invention relates to the technical field of big data, in particular to a real-time commodity sales prediction system and method based on edge calculation. The system comprises a data acquisition module, an edge calculation module, a commodity sales prediction module, an optimization feedback module and an intelligent scheduling module. The system collects sales, inventory, customer behaviors and environment data in real time through the Internet of Things, and utilizes a graph embedding technology to construct a dynamic relation graph and enhance features. Modeling is carried out based on a space-time diagram network algorithm, and accurate prediction of commodity sales volume is realized. Model parameters are optimized through reinforcement learning, inventory replenishment, a distribution path, a production schedule and a promotion strategy are dynamically adjusted in combination with a discrete differential evolution algorithm, and a real-time resource scheduling scheme is generated. According to the invention, the accuracy and real-time performance of commodity sales volume prediction are improved, optimal configuration of resources is realized through intelligent scheduling, and the operation efficiency and market competitiveness of enterprises are effectively improved.

The invention relates to the technical field of agricultural machinery task scheduling, in particular to a multi-target task optimization scheduling method based on agricultural machinery operation efficiency. The method comprises the following steps of performing agricultural machine operation efficiency parameter measurement on each agricultural machine under different operation conditions to obtain an agricultural machine operation efficiency data set; performing agricultural machine capability modeling according to the agricultural machine operation efficiency data set to obtain an agricultural machine capability matrix; the method comprises the following steps: partitioning a farmland by using a remote sensing technology to obtain a farmland partition map; performing farmland operation task decomposition according to the farmland partition map, and generating a sub-task set to obtain the sub-task set; and according to the agricultural machine capability matrix, cooperative edge generation is carried out, cooperative edge cost weighting processing is carried out, and a weighted cooperative edge set is obtained. According to the method, the balance of the scheduling scheme among multiple targets is realized by constructing the agricultural machinery capability matrix, the cooperation relation graph and the multi-target optimization model, and the disadvantage that the existing method is insufficient in consideration of agricultural machinery cooperation operation is effectively solved.

The invention discloses an enterprise big data security early warning method based on anomaly detection, and the method comprises the following steps: S1, collecting original data, and carrying out the format unification and structure standardization processing; s2, preprocessing is carried out, and feature vectors are constructed; s3, a behavior entity relation graph is constructed, a graph attention network is adopted for training, and structural features in a normal behavior mode are learned; s4, calculating the deviation degree between the current behavior and the normal behavior; s5, reconstructing the feature vector, measuring the deviation degree between the current behavior and the standard behavior distribution by using a mahalanobis distance, and calculating the posterior anomaly probability of the behavior through a Bayesian updating mechanism; and S6, evaluating the risk level of the current behavior according to the posterior anomaly probability, and generating a corresponding early warning event. According to the method, the graph attention network and the Bayesian self-coding technology are fused, enterprise behavior anomaly detection and grading early warning are achieved, and the method has the advantages of being high in recognition precision, high in self-adaption and timely in response.

The invention discloses a network space map surveying and mapping method and system based on multi-source data fusion, and the method comprises the steps: obtaining a multi-source data set in a unified format based on network flow data, equipment information data and geographic position data; obtaining a network asset entity and an incidence relation graph thereof through entity identification and correlation analysis based on the multi-source data set with the uniform format; obtaining a network space three-dimensional map model through three-dimensional space mapping and visual rendering based on the network asset entity and the association relationship map thereof; based on the network space three-dimensional map model, performing dynamic updating according to the accessed real-time data flow to obtain real-time network space situation data; and obtaining a network security risk assessment result through anomaly detection and threat identification based on the real-time network space situation data. According to the invention, visual display and security situation awareness of the network space are realized, and a brand new decision support tool is provided for network security management.

An approach for determining how well a graph neural network (“GNN”) learns a relation in a multi-relational graph (“MRG”) and generating a generalized relation to substitute in the MRG to more effectively train the GNN to learn the relation. Layers of interest in a GNN may be identified along with which parameters to monitor based on the architecture of the GNN and the model type of the MRG. A MRG may be input into the GNN to generate embeddings for the entities and relations of the MRG. A comprehension score can be generated for the relations and the relations can be ranked. A more generalized relation can be identified for relations ranked below a threshold and can be substituted into the multi-relation graph in place of the lower ranked more specific relations. The updated MRG can be input into the graph neural network to further train the GNN.

The invention relates to a BIM technology-based sponge city model construction method and system, and the method comprises the following steps: carrying out the classification of city underlying surfaces based on fused geographic information data, and obtaining underlying surface types and an underlying surface type topological relation graph corresponding to the underlying surface types; performing rainwater runoff path simulation analysis on the target sponge city based on the underlying surface type topological relation graph to obtain a rainwater runoff path network; based on the underlying surface type and the rainwater runoff path network, designing an urban rainwater regulation and storage facility layout of the target sponge city to obtain a rainwater regulation and storage facility layout scheme; and constructing a virtual sponge city model based on the fused geographic information data, the underlying surface type topological relation graph, the rainwater runoff path network and the rainwater regulation and storage facility layout scheme, thereby solving the problem that a traditional method is difficult to comprehensively consider complex and changeable urban landforms. The influence of different types of ground coverings on the rainwater runoff path is difficult to reflect.